Unstructured Road Detection and Steering Assist Based on HSV Color Space Segmentation for Autonomous Car

Abstract

One of the important things in a self-driving car (SDC), also known as an autonomous vehicle (AV) is detecting the road so that it remains in the right lane. Therefore this paper aims to be able to detect roads, especially unstructured roads based on the results of the HSV color space segmentation on the road, then produce car position information from the center of the lane (center offset) which is a parameter in making the decision to move the car's steering wheel to return to the center of the lane. In marking the edge of the roadside, the method used is Hough transform based on the resulting edge line using an edge detector, then the coordinates of the left and right curb lines which represent the width of the road. The results of this paper indicate that the system's ability to distinguish between road and non-road areas in several sections with an average percentage of 99.59% for accuracy, 99.49% for precision, and 98.84% for recall and the system's ability to mark the left and right edge of the roadside is very good with an average percentage reaches 99.27% and the percentage error and accuracy obtained in providing information on the position of the car from the center (center offset) based on the actual value and prediction results are 16.05% and 84.14%.